Registered: Mar 2007
Posts: 5

As part of a final project in 3D modeling, I decided to show my instructor a working CA implementation in MEL since he had been encouraging us to find new ways in which modeling software could be used(such as architecture, visualization, etc.). So far I have wrote a very simple program for running 1d elementary(binary) CA. I'm thinking about adding new rules(such as Margolus), support for totalistic CA, and adding the support for 2d and 3d CA as well. I've noticed that MEL doesn't seem to be the best language for implementing CA, but it's been interesting to try and I'm pleased with what I've accomplished so far. Any pointers as to what features I could add, or programming suggestions would be appreciated. I've attached some screenshots of the CA it produces...

Report this post to a moderator | IP: Logged

Registered: Mar 2007
Posts: 5

This is rule 30, n = 71, after 71 steps

Michael Colson has attached this image:

Report this post to a moderator | IP: Logged

Registered: Mar 2007
Posts: 5

This is rule 110, after 71 steps.

Michael Colson has attached this image:

Report this post to a moderator | IP: Logged

Registered: Mar 2004
Posts: 132

Hi Michael,

Your idea to implement CA's in Maya is totally awsome! I would love to see whatever animations you create for your final project.

A binary 27 cell 3D CA contains 2^27=134,217,728 individual definitions. I am not sure if this is reasonable for your means.

If you do not intend to browse all possible rules, and instead look only at a specific class, and excellent class with rich behavior are the totalistic rules. if your rules were totalistic, meaning that the cells are just averaged and the rules transform graylevel values, you would have just 27*(2-1)+1 = 28 definitions in this case. This class is much more manageable, includes all of the possible game of life rules, as they are just dependent counts of cells; they are always symmetric, which can be nice, and they have a higher ratio of complex behavior than the full space.

As for the implementation, I am unfamiliar with PERL, so I do not know what is reasonable to do. All you would need to do is find a good way to set the value of each cell to the sum of all of its neighbors plus itself. Then your definition table would be in the form of a 28 element list ranging from 0 to 27, and a binary digit associated with each one. Thus there are 134,217,728 totalistic rules here, instead of rule slots... : ). Then you can scan through the summed space and replace each of them with their respective values, then repeat the process as many times as you wish.

Here are a collection of 3D totalistic cellular automata for you. The rules are from 2 to 264 in steps of 4 — a total of 64 different rules.

Let us know if you need any more assistance. I'd be glad to help you out.

Jesse Nochella has attached this image:

Report this post to a moderator | IP: Logged